Experimentally increasing the rate of food consumption can disassociate satiety from the amount of food ingested, potentially leading to overeating. The Mandometer is a portable weighing scale connected to a small computer that can generate data recording detailed patterns of eating behaviour. The user puts a measured portion of food on the scale and the computer records and displays, in real time, the weight loss from the plate as the user eats: time zero on the graph effectively displays total portion size. This approach to parameterising an otherwise difficult to measure lifestyle factor has not been applied to the general population, to date only being used in therapeutic settings. Here we used the Mandometer to measure natural variation in eating behaviour in a random sample from an existing cohort study, The Avon Longitudinal Study of Parents and Children. In 91 samples assessed over three meals, those with complete data showed an average meal weight for normal weight category was estimated to be 275g (95% CI 251-298g). On average, normal weight participants consumed food at a rate of 0.24g per second (95% CI 0.23-0.24). There was evidence that overweight (p=0.01) and obese (p=0.01) participants had larger meals than normal weight participants on average. Those overweight participants on average had 102g more food than normal weight participants (95% CI 29-175g) and obese participants had 200g more food than normal weight participants (95% CI 41-358g). There was strong evidence that overweight and obese participants ate at a faster rate than normal weight participants. Overweight participants ate 0.2g per second quicker (95% CI 0.19-0.21g/s; p<0.005) than normal weight participants, while obese participants ate 0.56g per second faster on average (95% CI 0.52-0.60g/s; p<0.005). Furthermore, there was evidence that carriers of different genotypes at a confirmed fat mass and obesity related genetic locus (FTO) thought to act through patterns of consumption, showed different Mandometer results. Those carrying the T/T genotype on average had 130g less food than those carrying the rare A/A homozygote genotype (95% CI 57-204g) with a trend observed for the heterozygote state (45g less on average 95% CI (19-111g)). There was strong evidence that carriers of the common allele at rs9939609 (T) ate at a slower rate than normal weight participants. T/T participants ate 0.3g per second slower (95% CI 0.28-0.26g/s; p<0.001) than those with an A/A genotype. Of note, these patterns were correlated with pertinent health factors such as blood pressure measured at approximately the same age (~22 years). Taken together, this work highlights the ability of the Mandometer to be used as an effective measure of eating behaviour at the population level.